4k 28/36 VS. 3k 28/36

[Rick Darge]

http://www.redforrent.com/REDCODE.mov [h.264 960x639]
http://www.redforrent.com/CHARLIECODE.mov



I shot this candle, 8 different ways. I will try to do another test that isn't in the dark, something high detail, but this will give you an idea of what's going on with the new codec.

Workflow...

Shot on CF cards. [4k 2:1 / 3k 2:1 / Redcode 28/36- 500 ASA - Red Zoom f/2.8 - 24 fps]

Redcine to process to PRO-RES HQ 1080 at the BEST settings available.

Cut in FCP 6 - Canvas was set to 1920x1278 - PRO-RES HQ 1080 Output --->Into Compressor ---> h.264 compression 960x639 - Quality set to HIGH


F.Y.I. I did some duration tests since that's what matters to me most right now. I like going out with Red as light as possible, thus CF cards. The longer record times will matter out there.. This test was non-scientific.. I put the lens cap on/off the Red, recorded until the card filled up, and checked the length on a Mac

8GB CF RED CARD


LENS CAP ON ~ NO MOTION, NO DETAIL

3k 2:1 - RedCode 28 8m23s
3k 2:1 - RedCode 36 6m39s
4k 2:1 - RedCode 28 4m43s
4k 2:1 - RedCode 36 4m10s


LENS CAP OFF ~ RECORDING HIGH MOTION DETAIL

3k 2:1 - RedCode 28 8m23s
3k 2:1 - RedCode 36 6m44s
4k 2:1 - RedCode 28 4m43s
4k 2:1 - RedCode 36 4m13s

These record times make no sense to me.. I would think a scene with more detail would result in a higher data rate being used? Yet, when using Redcode 36 w/ 3k or 4k, you get 3-5 seconds of increased record time???? I have to retest..

 

[Andrew Walker]

No file.

 

[Laco Zamba]

Not Found

 

[Álex Montoya]

Recording black probably reduced the difference between 28 and 36.

Try a very detailed scene if you can, even if the camera is not moving, since REDCODE isn' intraframe.

 

[Ruairi Robinson]

http://www.redforrent.com/REDCODE.mov [h.264 960x639]



I shot this candle, 8 different ways. I will try to do another test that isn't in the dark, something high detail, but this will give you an idea of what's going on with the new codec.

Workflow...

Shot on CF cards. [4k 2:1 / 3k 2:1 / Redcode 28/36- 500 ASA - Red Zoom f/2.8 - 24 fps]

Redcine to process to PRO-RES HQ 1080 at the BEST settings available.

Cut in FCP 6 - Canvas was set to 1920x1278 - PRO-RES HQ 1080 Output --->Into Compressor ---> h.264 compression 960x639 - Quality set to HIGH


F.Y.I. I did some duration tests since that's what matters to me most right now. I like going out with Red as light as possible, thus CF cards. The longer record times will matter out there.. This test was non-scientific.. I put the lens cap on the Red, recorded until the card filled up, and checked the length on a Mac.. I don't know if recording black means less compression, etc...

8GB CF RED CARD

3k 2:1 - RedCode 28 8m23s
3k 2:1 - RedCode 36 6m39s
4k 2:1 - RedCode 28 4m43s
4k 2:1 - RedCode 36 4m10s

Yes, what you've shot here doesn't look like it'll test the limits of redcode, even at the lower data rate.

Perhaps you could try shooting something with intricate detailed textures - canvas, or grass. All in focus. That long lens stuff on a plain BG makes it really easy to compress...

also the highlights and flame are clipping - is this the grade, or is the detail lost?

Gotta protect the highlights! Makes it look a bit video-ish...

Best,
R.

 

[Rick Darge]

Recording black probably reduced the difference between 28 and 36.

Try a very detailed scene if you can, even if the camera is not moving, since REDCODE isn' intraframe.

I just did that. I set the camera up to record the full frame of a TV show playing off a 42" LCD.. Roughly the same record times except for Redcode-36 .....??

 

[Rick Darge]

I'll do some simple shots outside once light is out. I really want to see when to use these options and when not. I think its important to see dull objects and how they react to the code. Based on the candle footage, I would always shoot Redcode-28.. I don't see any differences in the 4k/3k raw files. I'm uploading a new test of the dog shot with a blue-gel 650w & a 30w incandescent bulb.

Yes, what you've shot here doesn't look like it'll test the limits of redcode, even at the lower data rate.

Perhaps you could try shooting something with intricate detailed textures - canvas, or grass. All in focus. That long lens stuff on a plain BG makes it really easy to compress...

also the highlights and flame are clipping - is this the grade, or is the detail lost?

Gotta protect the highlights! Makes it look a bit video-ish...

Best,
R.

 

[Rick Darge]

Here is another test,

Same parameters as the first one.

http://www.redforrent.com/CHARLIECODE.mov

 

[Kenn Michael]

You're a madman Richard! Go to bed!!!!

Wait, what am I still doing up?

 

[Rick Darge]

What's bed? Is that a new feature in Build 14?

 

[Kenn Michael]

Nope, the bed feature was intentionally left out of the RED product design. Bastards.

 

[Häakon]

Wow... that difference in FOV is pretty immense even with just the difference between 3K and 4K. Did you move any further away from the dog when you changed to 4K?

 

[Häakon]

I also must say... although I don't understand why the 4K record times between 28 and 36 are near identical (??), it is definitely nice to have the option of recording double the amount of footage on a card/drive in 3K. 3K should stand to be pretty nice for event work with a RED DRIVE for 1080p finishes.

 

[Rick Darge]

Haakon, I stood the same distance from the dog for each take. I'm still not seeing what Red-36 is doing. I need to go shoot grass in the morning..

 

[Laco Zamba]

28/36:

Maybe you need new REDCINE.
Or try new REDAlert and compare two identical frames pixel to pixel.

 

[Karl H]

Hi Richard,

Try exporting a DPX and look at the blue channel solo in photoshop. Generally thats where redcode has shown most of it's compression artifacts. Hopefully this is where the difference lies.

 

[Rick Darge]

i will take a look in the morning.. Redcine is tied up at the moment

 

[Mike Prevette]

It was stated in one of the threads that 28/36 was the MAX bitrate the compression would be capped at. So In theory if you were to shoot a bland low detail scene in both they could end up being the same. It's when you start shooting high frequency detailed scene that the difference will be noticed. Of course who knows what other magic Red has packed into it.

 

[BASSAM MSSALATIE]

Richard I think you are our man in the Lab .thanks for tests:usd:

 

[sparkhope]

I'm sorry to see that the conversation has stopped...has there been any advancements made in understanding the difference between rc28/rc36? Any evidence?

Thanks